[Telemetry] server side fetcher

[Bamieh]

Add a server side telemetry fetcher under a config flag (default: false). This enables any distribution to start sending telemetry from kibana server instead of browser.

[elasticmachine]

Pinging @elastic/kibana-stack-services (Team:Stack Services)

[stevewritescode]

Are there any kibana plugins planning to use this in 7.5?

[alexfrancoeur]

I don't believe there are any specific plugins planning to use this initially. Happy to touch base tomorrow to provide some more details @stevewritescode (awesome alias btw)

[Bamieh]

@stevewritescode just for clarification this is the fetcher method. (The part of telemetry that sends code back to our telemetry cluster) This is configured via kibana.yml executed only in the telemetey plugin.

From your question I understood that you're asking about telemetry server side usage collectors. If that is the case then we do already have server side collectors already being used by many plugins already.

[alexfrancoeur]

@Bamieh will we be sending duplicate documents if there are multiple kibana instances for a cluster? How are we handling that today?

[Bamieh]

@alexfrancoeur

Currently we get duplicate documents as each logged in user will send telemetry data once per day per browser.

• For the index xpack-phone-home we only have the latest document per cluster_uuid (no duplicates).
• For the indices all-xpack-phone-home-* we have duplicate documents.

Actually with the server side implementation we will have less duplicates since the number of users logging into kibana are much more than the number of kibana instances a cluser would have:

1. Our data indicates that most clusers have 1 kibana instance per cluser.
2. The most common case to add more kibana instances is due to having more users using kibana, so the number of instances will usually be less than the number of users using kibana.

This model works fine with the usage data we currenrly send so duplicates should not be an issue.

[alexfrancoeur]

++ thanks for the explanation @Bamieh. I had been under a similar impression as well (that we're capturing duplicates now and there will be less with server side). As always, appreciate the quick response

[mikecote]

Closed via #50015